Effectiveness of App-Based Cognitive Screening for Dementia by Lay Health Workers in Low Resource Settings. A Validation and Feasibility Study in Rural Tanzania

Authors:
Open Access
Publisher website
Google Scholar
Abstract
Background: The majority of people with dementia live in low-and middle-income countries (LMICs). In sub-Saharan Africa (SSA) human-resource shortages in mental health and geriatric medicine are well recognized. Use of technological solutions may improve access to diagnosis. We aimed to assess the diagnostic accuracy of a brief dementia screening mobile application (app) for non-specialist workers in rural Tanzania against blinded gold-standard diagnosis of DSM-5 dementia. The app includes 2 previously-validated culturally appropriate low-literacy screening tools for cognitive (IDEA cognitive screen) and functional impairment (abbreviated IDEA-IADL questionnaire). Methods: This was a 2-stage community-based door-to-door study. In Stage1, rural primary health workers approached all individuals aged ≥60 years for app-based dementia screening in 12 villages in Hai district, Kilimanjaro Tanzania. In Stage 2, a stratified sub-sample were clinically-assessed for dementia blind to app screening score. Assessment included clinical history, neurological and bedside cognitive assessment and collateral history. Results: 3011 (of 3122 eligible) older people consented to screening. Of these, 610 were evaluated in Stage 2. For the IDEA cognitive screen, the area under the receiver operating characteristic (AUROC) curve was 0.79 (95% CI 0.74-0.83) for DSM-5 dementia diagnosis (sensitivity 84.8%, specificity 58.4%). For those 358 (44%) completing the full app, AUROC was 0.78 for combined cognitive and informant-reported functional assessment. Conclusions: The pilot dementia screening app had good sensitivity but lacked specificity for dementia when administered by non-specialist rural community workers. This technological approach may be a promising way forward in low-resource settings, specialist onward referral may be prioritized.
Funding Information
  • National Institute for Health Research (16/137/62)

This publication has 31 references indexed in Scilit: